The question product teams keep getting wrong isn't how to make AI better. It's where to put it in the first place. Most teams make this decision at the product level ("we need an AI product") or the feature level ("add AI to search"). Both are too coarse. The right unit is the individual user story, the specific surface in your product where AI either belongs or doesn't. This talk introduces decision surfaces, a framework for mapping the concrete points in your product where AI can be implemented versus where it should stay out. Not every surface needs AI. Not every surface should avoid it. The framework helps you make that call precisely instead of defaulting to "AI everywhere" or retreating to "AI nowhere".